[lkml]   [2007]   [Apr]   [6]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [PATCH] Correctly prevent IDE timer expiry function to run if request was already handled

Suleiman Souhlal wrote:

>>>It is possible for the timer expiry function to run even though the
>>>request has already been handled: ide_timer_expiry() only checks that the
>>>handler is not NULL, but it is possible that we have handled a request
>>>(thus clearing the handler) and then started a new request
>>>(thus starting the timer again, and setting a handler).

>>>A simple way to exhibit this is to set the DMA timeout to 1 jiffy and
>>>run dd: The kernel will panic after a few minutes because
>>>ide_timer_expiry() tries to add a timer when it's already active.

>>>To fix this, we simply add a request generation count that gets
>>>incremented at every interrupt, and check in ide_timer_expiry() that we
>>>have not already handled a new interrupt before running the expiry

>>Couldn't this be addressed by simply changing add_timer() to mod_timer()?

> No, we don't want to run the expiry function at all, in this case, since
> the request might have correctly been handled already by the time we
> would try to run the expiry function/restart the timer.

> Also, if we just change the add_timer() to mod_timer(), we will just be
> hiding the problem because you might end up changing the timeout of a
> timer whose purpose is different (for a new request, for example).
> The timer should not be active when ide_timer_expiry() tries to restart
> it, since that function is called when the timer has expired (meaning it
> is not active anymore).

Yeah, that was stupid idea. Been looking at network schedulers too much
recently. :-)

> The reason the timer could have been active at that point, before applying
> this patch, is that we try to dispatch a new request after handling one.
> The new request will then have its own expiry timer, along with a handler.
> Since before this patch ide_timer_expiry() only looked at whether or not
> a handler was present, it would incorrectly think the request had not been
> handled already, and incorrectly tried to restart the timer.

Hm, I'm still not sure why this happens at all, probably need to try
reproducing it (unless you post a stack trace :-).

> -- Suleiman

MBR, Sergei
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
More majordomo info at
Please read the FAQ at

 \ /
  Last update: 2007-04-06 23:11    [W:0.049 / U:0.164 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site